diff --git a/swh/deposit/api/checks.py b/swh/deposit/api/checks.py index f76fba9c..dbf61404 100644 --- a/swh/deposit/api/checks.py +++ b/swh/deposit/api/checks.py @@ -1,129 +1,129 @@ # Copyright (C) 2017-2022 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information """Functional Metadata checks: Mandatory fields: - 'author' - 'name' or 'title' Suggested fields: - metadata-provenance """ import dataclasses import functools from typing import Dict, Optional, Tuple from xml.etree import ElementTree import pkg_resources import xmlschema from swh.deposit.errors import FORBIDDEN, DepositError from swh.deposit.utils import NAMESPACES, parse_swh_metadata_provenance MANDATORY_FIELDS_MISSING = "Mandatory fields are missing" INVALID_DATE_FORMAT = "Invalid date format" SUGGESTED_FIELDS_MISSING = "Suggested fields are missing" METADATA_PROVENANCE_KEY = "swh:metadata-provenance" @dataclasses.dataclass class Schemas: swh: xmlschema.XMLSchema11 codemeta: xmlschema.XMLSchema11 @functools.lru_cache(1) def schemas() -> Schemas: def load_xsd(name) -> xmlschema.XMLSchema11: return xmlschema.XMLSchema11( pkg_resources.resource_string("swh.deposit", f"xsd/{name}.xsd").decode() ) return Schemas(swh=load_xsd("swh"), codemeta=load_xsd("codemeta")) def check_metadata(metadata: ElementTree.Element) -> Tuple[bool, Optional[Dict]]: """Check metadata for mandatory field presence and date format. Args: metadata: Metadata dictionary to check Returns: tuple (status, error_detail): - (True, None) if metadata are ok and suggested fields are also present - (True, ) if metadata are ok but some suggestions are missing - (False, ) otherwise. """ suggested_fields = [] # at least one value per couple below is mandatory alternate_fields = { ("atom:name", "atom:title", "codemeta:name"): False, ("atom:author", "codemeta:author"): False, } for possible_names in alternate_fields: for possible_name in possible_names: if metadata.find(possible_name, namespaces=NAMESPACES) is not None: alternate_fields[possible_names] = True continue mandatory_result = [" or ".join(k) for k, v in alternate_fields.items() if not v] # provenance metadata is optional provenance_meta = parse_swh_metadata_provenance(metadata) if provenance_meta is None: suggested_fields = [ {"summary": SUGGESTED_FIELDS_MISSING, "fields": [METADATA_PROVENANCE_KEY]} ] if mandatory_result: detail = [{"summary": MANDATORY_FIELDS_MISSING, "fields": mandatory_result}] return False, {"metadata": detail + suggested_fields} deposit_elt = metadata.find("swh:deposit", namespaces=NAMESPACES) if deposit_elt: try: schemas().swh.validate(deposit_elt) except xmlschema.exceptions.XMLSchemaException as e: return False, {"metadata": [{"fields": ["swh:deposit"], "summary": str(e)}]} detail = [] for child in metadata: for schema_element in schemas().codemeta.root_elements: if child.tag in schema_element.name: break else: # Tag is not specified in the schema, don't validate it continue try: schemas().codemeta.validate(child) except xmlschema.exceptions.XMLSchemaException as e: detail.append({"fields": [schema_element.prefixed_name], "summary": str(e)}) if detail: - return False, {"metadata": detail} + return False, {"metadata": detail + suggested_fields} if suggested_fields: # it's fine but warn about missing suggested fields return True, {"metadata": suggested_fields} return True, None def check_url_match_provider(url: str, provider_url: str) -> None: """Check url matches the provider url. Raises DepositError in case of mismatch """ provider_url = provider_url.rstrip("/") + "/" if not url.startswith(provider_url): raise DepositError( FORBIDDEN, f"URL mismatch: {url} must start with {provider_url}", ) diff --git a/swh/deposit/tests/api/test_checks.py b/swh/deposit/tests/api/test_checks.py index 3cc96c0b..01ebd382 100644 --- a/swh/deposit/tests/api/test_checks.py +++ b/swh/deposit/tests/api/test_checks.py @@ -1,667 +1,712 @@ # Copyright (C) 2017-2022 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information # disable flake8 on this file because of line length # flake8: noqa import re import textwrap from typing import Any, Dict from xml.etree import ElementTree import pytest from swh.deposit.api.checks import ( METADATA_PROVENANCE_KEY, SUGGESTED_FIELDS_MISSING, check_metadata, ) METADATA_PROVENANCE_DICT: Dict[str, Any] = { "swh:deposit": { METADATA_PROVENANCE_KEY: {"schema:url": "some-metadata-provenance-url"} } } XMLNS = """xmlns="http://www.w3.org/2005/Atom" xmlns:swh="https://www.softwareheritage.org/schema/2018/deposit" xmlns:codemeta="https://doi.org/10.5063/SCHEMA/CODEMETA-2.0" xmlns:schema="http://schema.org/" """ PROVENANCE_XML = """ some-metadata-provenance-url """ _parameters1 = [ textwrap.dedent(metadata_ok) for (metadata_ok,) in [ ( f"""\ something something-else foo someone {PROVENANCE_XML} """, ), ( f"""\ something something-else foo no one {PROVENANCE_XML} """, ), ( f"""\ some url bar no one {PROVENANCE_XML} """, ), ( f"""\ some url bar someone should allow anything here {PROVENANCE_XML} """, ), ( f"""\ some url bar should allow anything here someone {PROVENANCE_XML} """, ), ( f"""\ some url bar should allow anything here someone should allow anything here {PROVENANCE_XML} """, ), ( f"""\ some url some id nar no one 2020-12-21 2020-12-21 2020-12-25 {PROVENANCE_XML} """, ), ( # technically, only Date is allowed for datePublished; but we allow DateTime # for backward compatibility with old swh-deposit versions f"""\ some url some id nar no one 2020-12-21T12:00:00 2020-12-21T12:00:00 2020-12-25T12:00:00 {PROVENANCE_XML} """, ), ( f"""\ something something-else bar someone """, ), ( f"""\ something something-else bar someone some-metadata-provenance-url """, ), ( f"""\ something something-else bar someone some-metadata-provenance-url """, ), ( f"""\ something something-else bar someone some-metadata-provenance-url """, ), ] ] @pytest.mark.parametrize( "metadata_ok", _parameters1, ) def test_api_checks_check_metadata_ok(metadata_ok, swh_checks_deposit): actual_check, detail = check_metadata(ElementTree.fromstring(metadata_ok)) assert actual_check is True, f"Unexpected result: {detail}" if "swh:deposit" in metadata_ok: # no missing suggested field assert detail is None else: # missing suggested field assert detail == { "metadata": [ { "fields": [METADATA_PROVENANCE_KEY], "summary": SUGGESTED_FIELDS_MISSING, } ] } _parameters2 = [ (textwrap.dedent(metadata_ko), expected_summary) for (metadata_ko, expected_summary) in [ ( f"""\ something something-else someone {PROVENANCE_XML} """, { "summary": "Mandatory fields are missing", "fields": ["atom:name or atom:title or codemeta:name"], }, ), ( f"""\ something something-else foobar {PROVENANCE_XML} """, { "summary": "Mandatory fields are missing", "fields": ["atom:author or codemeta:author"], }, ), ( f"""\ something something-else bar someone {PROVENANCE_XML} """, { "summary": "Mandatory fields are missing", "fields": ["atom:name or atom:title or codemeta:name"], }, ), ( f"""\ something something-else foobar foo {PROVENANCE_XML} """, { "summary": "Mandatory fields are missing", "fields": ["atom:author or codemeta:author"], }, ), ( f"""\ something something-else bar someone {PROVENANCE_XML} """, { "summary": "Mandatory fields are missing", "fields": ["atom:author or codemeta:author"], }, ), ] ] @pytest.mark.parametrize("metadata_ko,expected_summary", _parameters2) def test_api_checks_check_metadata_ko( metadata_ko, expected_summary, swh_checks_deposit ): actual_check, error_detail = check_metadata(ElementTree.fromstring(metadata_ko)) assert actual_check is False assert error_detail == {"metadata": [expected_summary]} _parameters3 = [ (textwrap.dedent(metadata_ko), expected_summary) for (metadata_ko, expected_summary) in [ ( f"""\ some url bar no one {PROVENANCE_XML} """, [ { "summary": ".*Reason: a simple content element can't have child elements.*", "fields": ["codemeta:name"], }, ], ), ( f"""\ some url bar no one {PROVENANCE_XML} """, [ { "summary": ".*Reason: character data between child elements.*", "fields": ["codemeta:author"], }, ], ), ( f"""\ some url bar should allow anything here {PROVENANCE_XML} """, [ { "summary": ".*Tag '?codemeta:name'? expected.*", "fields": ["codemeta:author"], }, ], ), ( f"""\ something something-else bar someone 2020-aa-21 2020-12-bb {PROVENANCE_XML} """, [ { "summary": ".*Reason: invalid value '2020-aa-21'.*", "fields": ["codemeta:datePublished"], }, { "summary": ".*Reason: invalid value '2020-12-bb'.*", "fields": ["codemeta:dateCreated"], }, ], ), ( f"""\ some url someid bar no one 2020-aa-21 2020-12-bb + {PROVENANCE_XML} """, [ { "summary": ".*Reason: invalid value '2020-aa-21'.*", "fields": ["codemeta:datePublished"], }, { "summary": ".*Reason: invalid value '2020-12-bb'.*", "fields": ["codemeta:dateCreated"], }, ], ), ( f"""\ some url someid bar no one 2020-12-aa + {PROVENANCE_XML} """, [ { "summary": ".*Reason: invalid value '2020-12-aa'.*", "fields": ["codemeta:dateModified"], }, ], ), + ( + f"""\ + + some url + bar + no one + + """, + [ + { + "summary": ".*Reason: character data between child elements.*", + "fields": ["codemeta:author"], + }, + { + "summary": "Suggested fields are missing", + "fields": ["swh:metadata-provenance"], + }, + ], + ), ( f"""\ something something-else bar someone + + some-metadata-provenance-url + """, [ { "summary": ( r".*Reason: Unexpected child with tag 'swh:invalid'.*" r"Instance:.*swh:invalid.*" ), "fields": ["swh:deposit"], } ], ), ( f"""\ something something-else bar someone + + some-metadata-provenance-url + """, [ { "summary": ( r".*Reason: Unexpected child with tag 'swh:add_to_origin'.*" ), "fields": ["swh:deposit"], } ], ), ( f"""\ something something-else bar someone + + some-metadata-provenance-url + """, [ { "summary": ( r".*Reason: assertion test if false.*" r"Schema:\n*" r' *]+ id="swhdeposit-incompatible-create-and-add".*' ), "fields": ["swh:deposit"], } ], ), ( f"""\ something something-else bar someone + + some-metadata-provenance-url + """, [ { "summary": ( r".*Reason: assertion test if false.*" r"Schema:\n*" r' *]+ id="swhdeposit-incompatible-create-and-reference".*' ), "fields": ["swh:deposit"], } ], ), ( f"""\ something something-else bar someone + + some-metadata-provenance-url + """, [ { "summary": ( r".*Reason: assertion test if false.*" r"Schema:\n*" r' *]+ id="swhdeposit-incompatible-add-and-reference".*' ), "fields": ["swh:deposit"], } ], ), ( f"""\ something something-else bar someone + + some-metadata-provenance-url + """, [ { "summary": r".*Reason: Unexpected child with tag 'swh:origin'.*", "fields": ["swh:deposit"], }, ], ), ( f"""\ something something-else bar someone + + some-metadata-provenance-url + """, [ { "summary": r".*Reason: Unexpected child with tag 'swh:origin'.*", "fields": ["swh:deposit"], }, ], ), ( f"""\ something something-else bar someone + + some-metadata-provenance-url + """, [ { "summary": r".*Reason: Unexpected child with tag 'swh:object'.*", "fields": ["swh:deposit"], }, ], ), ] ] @pytest.mark.parametrize("metadata_ko,expected_summaries", _parameters3) def test_api_checks_check_metadata_ko_schema( metadata_ko, expected_summaries, swh_checks_deposit ): actual_check, error_detail = check_metadata(ElementTree.fromstring(metadata_ko)) assert actual_check is False assert len(error_detail["metadata"]) == len(expected_summaries), error_detail[ "metadata" ] for (detail, expected_summary) in zip(error_detail["metadata"], expected_summaries): assert detail["fields"] == expected_summary["fields"] # xmlschema returns very detailed errors, we cannot reasonably test them # for equality summary = detail["summary"] assert re.match( expected_summary["summary"], summary, re.DOTALL ), f"Failed to match {expected_summary['summary']!r} with:\n{summary}"